NACA0012：Oscillation on the wall

General
22

I take a look in Gmsh as Nunn said.

I want to use your code to check grid，but it comes error，under the condition that h5py is installed. I use “python read.py” to launch the code.

23

Ok, I’m just a bit confused as the screenshot from @nnunn seems to show that although gmsh is using high order elements, the surface is faceted i.e. not high order.

24

Hi @jian - in PyFR, a high-order mesh implies that the mesh has curved elements. In your case, the triangles are all linear, i.e. all the edges are straight.

Your mesh is using 10-node triangles (high-order) to define triangles with straight edges. Usually, gmsh uses those 10-node triangles to define triangles with curved faces. So, to simulate a curved (high-order) aerofoil, we need to generate the gmsh .msh file using curved geometry (e.g. splines).

How did you generate your mesh? If you used gmsh, can you send the .geo “geometry” file?

25

Thanks a lot. This is my .geo file.

26

Hi @jian - using splines instead of lines through your points we get excellent curved geometry for gmsh to mesh. In this google folder you’ll find “NACA_p3_5751.msh”, a mesh with 5751 3rd order triangles, plus “NACA_curved.geo”, the geometry file used to generate the mesh.

If you want to reproduce and improve this mesh, I used these settings (with gmsh v4.9):

From the [Options] panel on the [Mesh/General] tab, I set:

  • 2D algorithm to [MeshAdapt]
  • Element order to [3]
  • Max element size [2]

Note that gmsh has lots of way to constrain the size of mesh elements. As a first step, I fiddled with “characteristic sizes” for the points to produce a mesh of size similar to yours. Better alternatives are to use background fields, boundary layers, nested transfinite regions, etc. See the gmsh documentation for inspiration.

Also in that folder are some pictures of the mesh, plus snapshots from various runs, each at time = 20:

[slip_wall_N4_t20.jpg]: Navier-Stokes / order 4 / mu 1e-6 / slip wall BC
[no_slip_wall_N4_t20.jpg]: Navier-Stokes / order 4 / mu 1e-6 / no-slip wall BC
[no_slip_wall_N4_t20_mu0.jpg]: Navier-Stokes / order 4 / mu 0 / no-slip wall BC

Good luck!

1 Like
27

Okey, thanks a lot.

Again, thanks for your kindness replay and best regards.